In recent years, robot learning has made impressive strides toward generally capable manipulation agents. Solutions powered by foundation models and extensive collections of human demonstrations have shown impressive generalization, while others based on pretrained vision/language models have exploited their common sense knowledge for diverse manipulation abilities. However, the demonstrated dexterity of these systems usually leaves one wanting: most works are limited to simple pick-and-place behaviors, far from the contact-rich skills robots need to accomplish many industrial and household tasks, from insertion and assembly to cooking and dexterous tool use. Meanwhile, recent model-based control and planning works have demonstrated impressive dexterity in contact-rich tasks, yet lack the generalization and scalability of data-driven approaches. This workshop brings together world-class experts from academia and industry to facilitate discussion around integrating these paradigms, with the goal of enabling robots to robustly perform diverse contact-rich tasks in the open world.
This workshop seeks to chart the course for the development of such manipulation behaviors beyond pick-and-place, building toward robots that can autonomously solve contact-rich manipulation problems in the open world, addressing the following questions:
How can we leverage model-based manipulation planning, task and motion planning, and optimal control together with learning-based methods for contact-rich manipulation?
How much can we learn about contact-rich interactions from images alone? What additional sensing modalities are required, and how can we scale data collection efforts? How should we use other sources, like human videos?
What are appropriate action spaces for contact-rich manipulation? How do we integrate in-hand and non-prehensile manipulation with prehensile abilities?
How can we enable safe, autonomous learning of contact-rich skills in the real world, through algorithmic and mechanical means?
09:00 - 09:15 Opening Remarks
09:15 - 09:45 Yifan Hou: Empower Robot Learning with Model-based Manipulation
09:45 - 10:15 Hae-Won Park: Contact-implicit Control and Estimation: Legged Robots and More
10:15 - 10:30 Paper Spotlights
10:30 - 11:00 Coffee + Posters
11:00 - 11:30 Ireti Akinola: A Simulation-First Approach to Robotic Assembly: Towards sensor-full contact-rich manipulation
11:30 - 12:00 Maria Bauza: Are we ready to go beyond pick and place?
12:00 - 13:00 Lunch Break
13:00 - 13:30 Katerina Fragkiadaki: 3D Generative Manipulation Policies and Object Dynamics
13:30 - 14:00 David Held: Spatially-aware Robot Manipulation
14:00 - 14:15 Paper Spotlights
14:15 - 14:45 Rachel Holladay
14:45 - 15:15 Michael Posa: Dexterity and generalization? A path toward to contact-rich model learning and control
15:15 - 15:30 Paper Spotlights
15:30 - 16:00 Coffee + Posters
16:00 - 16:30 Russ Tedrake: Multitask pretraining for dexterous, contact-rich manipulation
16:30 - 17:15 Debate / Panel: “Charting the Path Toward Contact-rich Manipulation”
featuring Russ Tedrake, Michael Posa, Rachel Holladay, Maria Bauza
17:15 - 17:30 Presentation of “Insights for Robot Learning Beyond Pick-and-Place”
17:30 - 17:45 Closing remarks, Best Paper award
See the venue on OpenReview here.
Sunshine Jiang, Xiaolin Fang, Nicholas Roy, Tomás Lozano-Pérez, Leslie Pack Kaelbling, Siddharth Ancha
DyWA: Dynamics-adaptive World Action Model for Generalizable Non-prehensile Manipulation
Jiangran Lyu, Ziming Li, Xuesong Shi, Chaoyi Xu, Yizhou Wang, He Wang
Tool-as-Interface: Learning Robot Tool Use from Human Play through Imitation Learning
Haonan Chen, Cheng Zhu, Yunzhu Li, Katherine Rose Driggs-Campbell
FoAR: Force-Aware Reactive Policy for Contact-Rich Robotic Manipulation
Zihao He, Hongjie Fang, Jingjing Chen, Hao-Shu Fang, Cewu Lu
GET-Zero: Graph Embodiment Transformer for Zero-shot Embodiment Generalization
Austin Patel, Shuran Song
Metric Semantic Manipulation-Enhanced Mapping via Belief Prediction Models
Nils Dengler, Joao Marcos Correia Marques, Jesper Mücke, Shenlong Wang, Kris Hauser, Maren Bennewitz
AugInsert: Learning Robust Visual-Force Policies via Data Augmentation for Object Assembly Tasks
Ryan Diaz, Adam Imdieke, Vivek Veeriah, Karthik Desingh
Robot Utility Models: General Policies for Zero-Shot Deployment in New Environments
Haritheja Etukuru, Norihito Naka, Zijin Hu, Seungjae Lee, Chris Paxton, Soumith Chintala, Lerrel Pinto, Nur Muhammad Mahi Shafiullah
Mobile Pedipulation for Object Sliding via a Wheeled Bipedal Robot
Yue Qin, Yanran Ding
SAIL: Faster-than-Demonstration Execution of Imitation Learning Policies
Nadun Ranawaka Arachchige, Zhenyang Chen, Wonsuhk Jung, Woo Chul Shin, Rohan Bansal, Yu Hang He, Yingyan Celine Lin, Benjamin Joffe, Shreyas Kousik, Danfei Xu
Learning Precise, Contact-Rich Manipulation through Uncalibrated Tactile Skins
Venkatesh Pattabiraman, Yifeng Cao, Siddhant Haldar, Lerrel Pinto, Raunaq Bhirangi
Reactive Diffusion Policy: Slow-Fast Visual-Tactile Policy Learning for Contact-Rich Manipulation
Han Xue, Jieji Ren, Wendi Chen, Gu Zhang, Fang Yuan, Guoying Gu, Huazhe Xu, Cewu Lu
Ben Abbatematteo
(UT Austin)
Roberto
Martín-Martín
(UT Austin)
Beomjoon Kim
(KAIST)
Harshit Khurana
(EPFL)
Aude Billard
(EPFL)
Jun Yamada
(Oxford)
Ingmar Posner
(Oxford)
Oliver Kroemer
(CMU)
Gentiane Venture
(UTokyo / AIST)